Tiger Kaovilai

🚀 Excited to announce the launch of Shamsi Calendar API! 🎉 After weeks of development, I'm proud to release a comprehensive RESTful API for the Afghan Shamsi (Solar Hijri) calendar - now live at https://lnkd.in/g7TNWXys Why I built this: As Afghan developers, we've long struggled with the lack of reliable tools for our calendar system. This API changes that. 🌟 Key Features: ✅ Accurate Shamsi ↔ Gregorian date conversion ✅ Full calendar views (day, week, month, year) ✅ Complete holiday management system ✅ Trilingual support: Dari, Pashto & English ✅ Interactive Swagger documentation ✅ 151 tests with 85.6% code coverage ✅ Free and open-source 🛠️ Built with: TypeScript, Express.js, tested with Jest, and deployed with enterprise-grade security (rate limiting, CORS, Helmet) 📊 What you can do: Build Afghan calendar apps without reinventing the wheel Integrate Shamsi dates into your existing applications Access accurate holiday information Use it in any language: TypeScript, Python, Java, etc. 🎯 Perfect for: Afghan tech startups, government systems, educational platforms, mobile apps, and anyone serving the Afghan community worldwide. Try it out at the interactive documentation and let me know what you think! Contributions and feedback are welcome. #OpenSource #API #AfghanTech #ShamsiCalendar #WebDevelopment #TypeScript #SoftwareEngineering #Afghanistan

22

1 Comment

This is my recent deep dive into LLM models. The "Fundamentals of Large Language Models" course provides a high-level overview of LLM architecture, scaling theory, and the current industry landscape. Here is a summary of the key themes covered: 1. Transformer Architectures 2. Embeddings vs. LLMs 3. Scaling Laws and the "Chinchilla" Shift 4. Benchmarking and "Openness" 5. Tokenization and embedding generation using tiktoken and cohere python packages. A more detailed version of this summary is available at https://lnkd.in/gvk_erme

1

If you’ve ever looked at the last 12 months of coding agents and thought, “Wait… what exactly is my job shifting into?” this Engineer’s Codex piece puts a clean frame around it: as implementation gets cheaper, engineering work moves “up the stack.” The provocative line is the title: everyone is a staff engineer now. Not because titles changed overnight, but because the skills we used to associate with staff+ (architecture, context management, system-level judgment) are becoming baseline expectations earlier. A few takeaways I’m sitting with: - Agents can refactor code, pass tests, and still break reality, especially when the system relies on undocumented “implicit contracts” living in team memory. - If code generation gets cheap, then reading, evaluating, and stewarding changes becomes the high-leverage work. - Long-running, asynchronous agent workflows are productive… until they quietly destroy flow through constant context switching. - The engineers who win won’t be the “best prompters.” They’ll be the ones who can define boundaries, decompose work, and steer execution with precision. For educators (and mentors), there’s an immediate implication: we should teach students earlier how to reason about systems, write and evaluate design decisions, and review code thoughtfully, not just produce it. #SoftwareEngineering #AI #DeveloperProductivity #EngineeringLeadership #SystemsThinking #CodeReview #SWE #ComputerScience #Engineering #DataScience #GenAI #AgenticAI #Agents #Coding #Programming

3

1 Comment

Retrieval is quietly becoming one of the most critical challenges in the AI landscape, especially as generative models transform how we interact with information. ZeroEntropy, a startup co-founded by Ghita Houir Alami, recently raised $4.2 million to tackle this problem head-on by improving how AI systems retrieve relevant data from messy, real-world knowledge bases. Their proprietary re-ranker, ze-rank-1, outperforms models from Cohere and Salesforce, ensuring AI doesn't just pull information—it pulls the right information first. What’s striking here is how foundational this layer is for AI’s accuracy. Many teams either hack together different search tools or dump entire knowledge bases into a model—approaches that quickly lead to errors and inefficiencies. ZeroEntropy offers an API designed specifically for developers to build robust retrieval systems that handle complex data at scale. Beyond the tech, Ghita’s journey from Morocco to launching a cutting-edge AI infrastructure startup highlights the importance of diverse perspectives tackling these hard problems. How are you approaching the balance between model capability and the reliability of the data those models access? What strategies have you found effective in reducing errors in AI-driven retrieval? #AIInfrastructure #RetrievalAugmentedGeneration #MachineLearning #StartupInnovation

1

1 Comment

this is actually insane (in the best way) Elijah Ghossein and Ali Hamad saw how research collaboration is and instead of complaining, they built something better. no code. no gatekeeping. just execution. that’s the kind of energy that stands out. lowkey the biggest takeaway: → you don’t need perfect tools, you need a clear problem → builders > observers → the barrier to creating real impact is basically gone decoded brain feels like one of those early signals 👀 Notion

1

I just published a new YouTube tutorial explaining LangGraph State Schemas, MessagesState, and Reducers. In this video, I break down how state works in LangGraph, starting from a simple schema with a single key and gradually moving to complex schemas with multiple keys and custom types. I also explain: • Why reducers are necessary to prevent state overwrites • How to annotate schema keys with reducers • How MessagesState works • How the add_messages reducer behaves • How to design a custom MessagesState schema for your own agents If you’re building LLM agents, chatbots, or multi-node workflows, this will help you understand how to manage state correctly in LangGraph. 🎥 Watch here: https://lnkd.in/gABnBiRt

5

From this week's episode of #Overcommitted, engineers with 2-3 years of experience hit the peak of the Dunning-Kruger Effect. Bassem Dghaidi is designing a system design course targeting exactly that demographic, helping engineers discover gaps in their knowledge and learn from others mistakes. Check out this week's episode for more insights from Bassem on technical growth and continuous learning. #SoftwareEngineering #SystemDesign #DeveloperEducation

25

3 Comments

My talk from AUC's Dev World is now up: https://lnkd.in/gaBNiEsV Some developers just seem to fly when things get hectic — when bugs pop up, deadlines tighten, or priorities shift overnight. What’s their secret sauce? In this talk, I share the habits, mindsets, and workflows that help you move fast without cutting corners or sacrificing quality. You’ll learn how to shorten your debug loop, master your IDE, terminal tricks, and build a mental map of your team’s work to avoid reinventing the wheel. I explore why sometimes slowing down to really understand a problem can actually speed you up, and how small automations and smart documentation can multiply your impact. Whether you’re at a scrappy startup or a big company, junior or senior, this talk is for anyone who wants to get sharper, stay calm under pressure, and deliver better, faster. If you have any more tips, I'd love to hear them.

11

1 Comment

I just published Part 2 of my series inspired by Andrej Karpathy’s idea: 👉 “LLMs are becoming the new Operating System.” In Part 1, I shared my journey through Windows, laptops, Android tablets, and how LLMs feel like the next massive shift in how we interact with computers. Today in Part 2, I’m opening up the hood and breaking down the actual architecture of the LLM OS. Karpathy visualized it with a simple diagram… and the deeper I looked at it, the more it felt like a blueprint of a new kind of computer. In this article, I break down: ⚙️ LLM as the CPU 🧠 Context window as RAM 💾 File system + embeddings as the disk 🧮 Classical tools as system calls 🎥 Video/Audio as peripheral I/O 🌐 Browsers + other LLMs as networking Everything we know from traditional computer systems is reappearing — but rebuilt around language, reasoning, and models instead of code and circuits. If you’re curious how an LLM actually functions like a computer, this part makes the picture clear. 📌 Read Part 2 here: https://lnkd.in/gDggQc52 Part 3 drops next: The Context Window — Why It’s the RAM of the LLM OS. Would love to hear your thoughts! 🚀

27

2 Comments